In this paper, compressive sensing (CS) was implemented to reconstruct the particle sizes in the particle sizing instrument with a digital micro-mirror device (DMD)-based detector. Different size distributions were used to compare the performance of the proposed sampling method. l(1)-norm-based reconstruction algorithm was implemented to retrieve the size distribution of particles. In the simulations, data contaminated by different additive noises were applied to size retrieval at several sampling rates. Experiment on a phantom was carried out in this work. The results showed that the proposed method based on CS was robust to the noises and efficient as less samplings were required. (C) 2016 Elsevier B.V. All rights reserved.